Ask about the day, not the strategy
The most common failure in an executive interview is to spend it on ambition. Ask a chief executive what the company will do differently and the answer is a direction of travel that applies to every company in the sector. Ask instead what happened last week, and the answer is specific, and it usually contains the opportunity. Executive discovering is an exercise in moving the conversation from intention to operation, without the interviewer appearing to check the executive's competence.
The working rule is that the interviewer should leave with specific episodes, named processes and repeated activities, and the executive should leave with a sharper view of their own operation than they had when the conversation started. Both are possible when the questions are about work rather than about strategy.
Questions that work across every role
- Which part of your week is a task you would happily not do again, and how much of it repeats? Repetition plus low attachment is the most reliable signal in discovery.
- Where does a decision get made that you would rather were made earlier, and what information would it need? This surfaces missing context rather than missing effort.
- What is the last thing your team stopped doing because it was not worth the effort? A process that was quietly abandoned tells you about a problem and about a workaround that may still exist elsewhere.
- How many times a month does something go wrong in a way that costs more than the time to fix? This converts a vague worry into a countable event.
Role-specific questions that reach the real constraint
Each executive sees a different part of the machine, and the questions below are written to reach the constraint that only that role can describe.
| Role | Question to ask | What the answer reveals |
|---|---|---|
| Chief executive | Which decision do you make with less information than you would like, and how long do you wait for it? | Where information exists but arrives too late to act on. |
| Chief operating officer | Which part of the operation cannot be handed to someone else without your judgement? | Where the organisation is short of operational capacity. |
| Chief financial officer | Where do you find out about a problem after it has cost money rather than before? | Which control failures are detected late or not at all. |
| Chief human resources officer | Which policy question consumes the most experienced judgement in your team? | Where expertise is concentrated in repeatable case work. |
| Department head | What does your team do every week that you would not include in a job description? | The administrative load the structure does not name. |
| Frontline manager | What do you spend the most time chasing, and who could do it instead? | The handoffs that carry the operational cost. |
How to interview without triggering a defence
Executives become guarded when an interview feels like an audit, and guarded answers are useless. Three things reduce it. Ask about the week rather than the year, because a week is concrete and harmless. Ask for the last occurrence rather than the general rule, because a story is easier to tell than a policy. And say plainly what the answers will be used for, and what will not be done with them, because an executive who is unsure of the purpose will give you the answer they think you want.
There is a fourth technique that is worth more than it sounds: ask the same question of two people who do the same work. Where the answers diverge, the organisation has two versions of the process, and the difference between them is where the rework lives. Nothing else in discovery surfaces that so cheaply.
From interview notes to a candidate list
Raw interview notes are not an opportunity list. The conversion is a deliberate act of sorting, and it should happen within a few days while the conversations are still fresh.
- Extract every repeated activity, and count how many roles mentioned it. Mentioned by several people it is probably real; mentioned by one it may be a personal workaround worth investigating separately. For each, note the baseline a change would show in: volume, elapsed time, or rework.
- For each candidate, name the person who carries the cost of it. If no one can be named, the candidate has no owner and it will not survive contact with a budget.
- Separate the candidates into work that is a rule, work that is judgement, and work that is a handoff. The three categories have different answers, and mixing them is the most common way an opportunity list produces a project nobody can deliver.
- Note the constraint the executive stated. Their constraint is a reliable predictor of what will be rejected later, and it should be written next to the candidate.
The output of a round of interviews is therefore a short list of processes with counts, owners and constraints attached, a baseline for each where one can be observed, and a record of the questions that remain open. One executive, typically the chief operating officer, should own that list and the decision that follows it, because a list without an owner is a document that is read once and discarded.
One caution belongs alongside the list. Some candidates are not AI opportunities at all: where the answer is a rule, a threshold or a reallocation of work, the correct output is a decision about the process, and a proposal that installs a model where the business needed a rule has made the problem harder, not easier.
The work that follows the list is described in finding your first ten high-value AI opportunities, and the interviews only work if the people running them are prepared to be told that the opportunity they expected is not the one that matters. More on the conditions for that conversation in why enterprise AI requires people who know the business from the inside. The engagement structure that follows from this is described in our transformation practice.